Accurate CT numbers obtained from Dual-energycomputed tomography (DECT), may improve the accuracy of radiation treatment planning. Our Dual energy Iterative Reconstruction Algorithm (DIRA) utilizes material decompositionin the iterative loop to produce accurate CT numbers of the imaged object. DIRA produces two mono-energy images. These mono-energies correspond to the effective energies of the two X-ray spectra. In simulations, the resulting CT numbers in the reconstructed images have shown to agree well with tabulated values. The aim of this work is to see if DIRA can also be used on real data. Compared to simulated data, the following adjustments were needed; i) Elimination of patient table. ii) Bow-tie filter modeling and adaptation to this in DIRA. iii) Application of inverse water beam hardening correction. A Siemens SOMATOM Force scanner was used to scan a cylindrical PMMA phantom with four rod inserts made of different materials. For simlicity, an axial scan was used. 6 iterations of DIRA was applied, where the 0th iteration is our own implementation of filtered backprojection (FBP). DIRA at iteration 6 gave values close to the tabulated ones. DIRA also performed well on an anthropomorphic head phantom (CIRS model 731-HN), with specified material composition.